Model Predictive Control of an Active Magnetic Bearing Suspended Flywheel Energy Storage System

Flywheel Energy Storage (FES) is rapidly becoming an attractive enabling technology in power systems requiring energy storage. This is mainly due to the rapid advances made in Active Magnetic Bearing (AMB) technology. The use of AMBs in FES systems results in a drastic increase in their efficiency. Another key component of a flywheel system is the control strategy. In the past, decentralised control strategies implementing PID control, proved very effective and robust. In this paper, the performance of an advanced centralised control strategy namely, Model Predictive Control (MPC) is investigated. It is an optimal Multiple-Input and Multiple-Output (MIMO) control strategy that utilises a system model and an optimisation algorithm to determine the optimal control law. A first principle state space model is derived for the purpose of the MPC control strategy. The designed MPC controller is evaluated both in simulation and experimentally at a low operating speed as a proof of concept. The experimental and simulated results are compared by means of a sensitivity analysis. The controller showed good performance, however further improvements need to be made in order to sustain good performance and stability at higher speeds. In this paper advantages of incorporating a system model in a model-based strategy such as MPC are illustrated. MPC also allows for incorporating system and control constraints into the control methodology allowing for better efficiency and reliability capabilities.